Clinical Epidemiology in the World of COVID: Sticking to the Basics
November 24. 2020
Covid-19 has raised awareness of the value of epidemiology in the eyes of society in general. Clinical epidemiology is a discipline within the larger discipline of epidemiology that is valuable for determining which diagnostic tests are most valuable, and which treatments are best for a given condition.
Covid-19 provides examples of where clinical epidemiology has substantial relevance. One example is with the development of diagnostic tests. Another is determining which treatments are “best” for a given patient. Clinical epidemiology provides a path for cutting through the hype and for developing testing and clinical products that are of value in real-world settings.
A starting point for using clinical epidemiology for diagnostic test development is determining test accuracy. The relevant concepts are test sensitivity (the true positives) and test specificity (the true negatives). While these concepts were once believed to be inherent characteristics of a given diagnostic test, the reality is that these characteristics can vary. For instance, a cardiac stress test is likely to correctly identify elderly patients with heart disease, but a “positive” in a 20-year-old patient is more likely to be a false positive.
The rush to develop diagnostic tests for Covid-19 illustrates the importance of understanding the fine points of test accuracy. While it is very lucrative to develop an accurate diagnostic test, the characteristics of a diagnostic test under laboratory conditions often don’t match what is observed in the field. There are a number of potential reasons why this occurs, including:
- Test development: diagnostic tests are often developed based on samples from individuals with marked clinical disease compared with samples from true negatives. But what about those in the early stages of disease? These individuals are often the ones that are of uncertain status in clinical settings. Or, if you are looking at big picture disease control concerns, how does the test perform for identifying individuals before they start infecting others? Thinking these considerations through can be the difference between developing a successful test and one that flops.
- What are the questions of concern? As was just alluded to, test interpretation is often considered at the level of diagnosing individuals. However, test interpretation can be different for addressing population level questions. If we are dealing with an infectious disease, there are multiple questions that can exist for individuals, populations, or combinations of the two:
- Are we concerned with knowing if someone is shedding an infectious agent, whether they are currently infected, whether they were ever infected, what their risk of transmitting or being exposed to the infectious agent is, or whether a rare disease is present in a population?
- These are different questions that might require the use of different tests or different interpretations of the same test. From a commercial perspective, directions for interpreting a test may vary depending on the particular clinical signs or population of interest.
- What is the prevalence of disease in a population? Population prevalence influences the accuracy of test interpretation. Thus, this is one of the moving targets that complicates test interpretation, but makes sense when you think about it. For instance, if a disease was very uncommon in a population, a positive test result could likely be a false positive. Similarly, if a disease is very common in a population, a negative test result could likely be a false negative. Understanding these relationships can be valuable for communicating with users about your product.
- Cross-reactions: In the case of Covid-19, will common corona viruses that cause GI disease in the young or the common cold be detected on tests that are intended for Covid-19. These questions and other cross-reactions are common concerns for infectious disease testing.
- There are many other potential test development and interpretation concerns that exist. While this does not invalidate the test results in any way, it does behoove one to understand clinical epidemiology with sufficient clarity that the strengths and weaknesses of different tests for a given use can be rigorously understood.
Extending these concepts further to treatment, which medications are of value? If there are public claims of efficacy, understanding clinical epidemiology can provide an objective means of evaluating your product or those of competitors. Recent announcements of pending approvals for Covid-19 vaccines from different manufacturers and the debate surrounding the efficacy of these vaccines in different populations highlight the challenges of product approval and public acceptance. As with diagnostic tests, there is a need to understand how the medication was evaluated. What age and stage of infection was assessed? Was the study prospective and were all participants consistently evaluated on the same criteria? What were those criteria and were all participants monitored through to a logical endpoint, or were there drop-outs?
Of course, the core question is what is the objective of treatment? Are you looking for a cure, or is the goal just symptomatic relief? Do you even need a treatment? Is the product cost-effective and if yes, when do you stop treatment? Thus, as with diagnostic tests, clarity and critical evaluation are important.
Clinical epidemiology provides a framework for addressing the above questions. While the above questions present a lot of uncertainty and variables to consider, this obviously does not mean that diagnostic tests and treatments are not of value. It does mean that there is a need to rigorously nail down your concerns for a given situation and ensure your understanding of the variables, such as stage of infection, subject age, and many other considerations. These details are relevant to product development, communication about a product, and efforts at continual improvement that ensure product relevancy.